The Selection of Acceleration Factors for Improving Stability of Particle Swarm Optimization
Wei Zhang, Hua Li, Zhaoxia Zhang, Wang Hua-kui · 2008
In this paper, the effect of acceleration factors on position expectation and variance in particle swarm optimization algorithm was studied. After statistic discuss in theory, a new parameter selection that setting the cognitive acceleration factor as 1.85 and the social acceleration factor as 2 has been proposed at the view of improving system stability. Five benchmark functions were used to test its efficiency comparing with the parameter selection that Kennedy was proposed that setting both of acceleration factors as 2. Numerous experiments and statistical results yield the efficiency of the new parameter selection which is beneficial to engineering application.